Object Recognition & Classification model based on Geometric Shapes in Virtual Environment
Budget: $750 – $1,500 USD
Object classification in virtual environment based on geometric shape recognition and comparison. This is NOT a data training or neural networks project
Below is a summary to do list, more detailed project description is also available, pls do not post a reply before reading it
1) Create a basic artificial environment; cube shaped room and/or display pedestal that has specific dimensions and grid like coordinates throughout its height, length, and width
2) Import 3D Objects from a free 3D Object library like sketch-up
2) Employ 3D geometric shapes; Cube, Sphere, Tetrahedron, Cuboid Rectangular Prism, Cone, Triangular Prism…. (8 to 12 shapes in total) and combine these shapes to overlay/replace 3D object with least number of geometric shapes possible and record original type for each utilized geometric shape (larger 2D area determines 3D geometric object type) and within pre-set ranges geometric shape dimensions can be scaled up or down, but they have to maintain primary characteristic of their original shapes
Suggestion for 3D Object to Geometric Shape conversion:
Using grid structure in virtual display platform, identify the center of volume/mess for the 3D Object
and start testing potential fit for different geometric shapes on that point.
For example: Start testing with a very small Triangular Prism placed at the center of volume point, expend geometric shape till it intersects with 3D Object and measure volume of 3D object covered that's within
closest layer of grid and measure geometric shape volume that's outside 3D object and closest layer of grid
and find the geometric shape with max 1st and min 2nd volume difference and once you identify best match
then identify areas of 3D object that lays outside the first geometric shape and their intersection points. Identify center of volume for each remaining part and apply different geometric shapes starting from their center of volume till they intersect with 3D Object and first geometric shape
3) Store following data for each revised 3D object:
a) List of Unique Geometric shapes utilized with original type names
b) Area of Geometric Shape relative to the artificial environment
c) Volume of Geometric Shape relative to the artificial environment
d) Length of Geometric Shape (longest dimension among width, length, and height)
e) Largest Geometric shape by volume
Compare geometric shapes of same category to each other based on width, length and height and compare geometric shapes to largest geometric shape
f) Intersection parameters of each geometric shape relative to each other, to largest shape and to artificial environment; angle, position, and area of intersection
For each geometric shape; count of intersections with other shapes, largest shape and display room
Position of intersection relative to center of 3D shape
Area of intersection as (%) of total shape area
Angle of intersection among shapes and display area relative to display platform
5) New software should have the ability to check match levels based on above attributes and create and name categories and subcategories of objects
Object classification process will proceed according to the following sequence:
a) Identification of Unique 3D Geometric Shape Types and largest geometric shape by volume
b) Identification of Total Number of shapes from each Unique Geometric Shape Type
c) Size measurement for each Geometric Shape; Area, Volume and Length
d) Comparison of geometric shapes within same category based on Area, Volume and Length against combined Average Values for all shapes in same category and to largest geometric shape (based on volume)
e) Identification of intersection points for each geometric shape with other shapes; count, position, area & angle of intersection with all other geometric shapes
I have more detailed information if you are truly interested and confident you can do it
Project Budget: $1,400
Below is a summary to do list, more detailed project description is also available, pls do not post a reply before reading it
1) Create a basic artificial environment; cube shaped room and/or display pedestal that has specific dimensions and grid like coordinates throughout its height, length, and width
2) Import 3D Objects from a free 3D Object library like sketch-up
2) Employ 3D geometric shapes; Cube, Sphere, Tetrahedron, Cuboid Rectangular Prism, Cone, Triangular Prism…. (8 to 12 shapes in total) and combine these shapes to overlay/replace 3D object with least number of geometric shapes possible and record original type for each utilized geometric shape (larger 2D area determines 3D geometric object type) and within pre-set ranges geometric shape dimensions can be scaled up or down, but they have to maintain primary characteristic of their original shapes
Suggestion for 3D Object to Geometric Shape conversion:
Using grid structure in virtual display platform, identify the center of volume/mess for the 3D Object
and start testing potential fit for different geometric shapes on that point.
For example: Start testing with a very small Triangular Prism placed at the center of volume point, expend geometric shape till it intersects with 3D Object and measure volume of 3D object covered that's within
closest layer of grid and measure geometric shape volume that's outside 3D object and closest layer of grid
and find the geometric shape with max 1st and min 2nd volume difference and once you identify best match
then identify areas of 3D object that lays outside the first geometric shape and their intersection points. Identify center of volume for each remaining part and apply different geometric shapes starting from their center of volume till they intersect with 3D Object and first geometric shape
3) Store following data for each revised 3D object:
a) List of Unique Geometric shapes utilized with original type names
b) Area of Geometric Shape relative to the artificial environment
c) Volume of Geometric Shape relative to the artificial environment
d) Length of Geometric Shape (longest dimension among width, length, and height)
e) Largest Geometric shape by volume
Compare geometric shapes of same category to each other based on width, length and height and compare geometric shapes to largest geometric shape
f) Intersection parameters of each geometric shape relative to each other, to largest shape and to artificial environment; angle, position, and area of intersection
For each geometric shape; count of intersections with other shapes, largest shape and display room
Position of intersection relative to center of 3D shape
Area of intersection as (%) of total shape area
Angle of intersection among shapes and display area relative to display platform
5) New software should have the ability to check match levels based on above attributes and create and name categories and subcategories of objects
Object classification process will proceed according to the following sequence:
a) Identification of Unique 3D Geometric Shape Types and largest geometric shape by volume
b) Identification of Total Number of shapes from each Unique Geometric Shape Type
c) Size measurement for each Geometric Shape; Area, Volume and Length
d) Comparison of geometric shapes within same category based on Area, Volume and Length against combined Average Values for all shapes in same category and to largest geometric shape (based on volume)
e) Identification of intersection points for each geometric shape with other shapes; count, position, area & angle of intersection with all other geometric shapes
I have more detailed information if you are truly interested and confident you can do it
Project Budget: $1,400